176 ON THE COLOURS OF THICK PLATES.
any part of the system. It was easy to know when, the image of the eye in the inclined plate lay in a line drawn through the luminous point perpendicular to the plane mirror, by observing when the image of the luminous point seen by reflexion, first at the plane mirror, and then at the plate of glass, lay in the centre of the system of rings. Supposing the image to have this position, on moving the head sideways the opening out of the rings could be traced from its very commencement. By moving back the head so as to keep the image of the luminous point in the centre of the system of rings, it was easy to try the experiment to which Art. 13 relates, the virtual image of the eye being thus kept in a line drawn through the luminous point perpendicular to the mirror, and the eye moving relatively to the luminous point, which is as good as if the luminous point had moved while the eye remained fixed. I found, in fact, that on moving back the head the rings expanded till the bright central patch surrounding the image filled the whole field of view, and on continuing to move back the head the rings appeared again. In the position in which the central patch filled the whole field of view, the least motion of the eye sideways was sufficient to bring into the field portions of excessively broad coloured bands.
Between the system of rings seen when the eye was respectively nearer to and further from the mirror than when in the position in which the rings became infinite, there was one difference which may here be mentioned. So long as the image occupied the centre of the systems, they were similar to each other; but when the head was moved sideways, the centre of the circles passed in the first case to the side of the image towards which the head was moved, and in the second case to the contrary side. This affords another way of comparing with experiment the result of theory already mentioned relating to the direction of curvature, arid it will be readily seen that the result of experiment agrees with the prediction of theory. For, suppose the distance of the eye from the oblique plate less than that of the luminous point, so that the virtual image of the eye lies between the luminous point and the mirror, and let the eye move to the right. Then its virtual image moves to the left, and therefore, according to theory, the centre of curvature ought to fall to the left of the image of the luminous point, right and left being estimated with reference to an eye